Google Adds Live Avatar to Gemini 3.8 Live

Google has officially launched Gemini 3.8 Live with a Live Avatar feature, bringing near real-time visual presence to its native real-time conversation model. The feature combines video generation with voice, using precise lip sync, natural expressions, and smooth turn-taking to create an interactive experience that listens, observes, and communicates through a dynamic visual avatar, helping businesses expand virtual services.

Live Avatar relies on Gemini's advanced reasoning and supports asynchronous tool calls, fetching data in the background while keeping conversations uninterrupted. It supports native multilingual voice sync and can switch seamlessly between 97 languages without losing video fidelity. Beyond preset avatar libraries, businesses can create custom avatars from high-quality reference images, preserving brand style and character traits, though this is currently limited to enterprise whitelist users.

All AI-generated content is embedded with invisible SynthID watermarks to help identify AI content and reduce misinformation. The feature is now available on Gemini Enterprise, and users can explore the API through official documentation.